I sometimes wonder about being able to fully simulate a human brain. Maybe even scan/copy a real person’s brain.
So many philosophical, ethical and legal questions. And unsettling possibilities.
We will probably have to deal with this someday.
> We will probably have to deal with this someday.
This is quite an extraordinary claim with no extraordinary evidence.
As said elsewhere in this thread we can at this moment not even simulate single atoms.
I see no reason to believe at all that we will ever be able to simulate a human brain.
Unless you want my simulation here:
if ishungry:
Eat(FindFood())
else:
PracticeFingersnowboarding()Why do you assume that we need to exactly simulate single atoms to simulate the brain? We can also do CFD to simulate fluid flows without simulating every atom in the fluid. Maybe that's possible with the brain, too.
Yes, and CFD simulations notoriously break down the moment you scale them up. Look at weather forecasting! Extremely complex models that push the boundaries of modern computational capabilities, and yet we are barely able to forecast weather a day or two out. Sure, we can make vague forecasts over large areas, but errors compound and we lose both granularity and accuracy.
And that's with us having a pretty solid understanding of how fluid dynamics works. We have an extremely poor understanding of how a brain works, doubly something of the complexity of the human brain. We are fundamentally unable to study it during operation, because we don't have a non-invasive high resolution access to it's internals. We are basically butchers sticking electrodes into living tissue.
The article itself proposes that we may - barely - be able to study the workings of the brain of an extremely simple organism.
A rocketry analogy would be Archimedes dreaming about people traveling to the stars.
I would say the issue with predicting weather by simulating flows isn’t mainly that we can’t simulate single atoms, it’s that we just don’t have enough input data. If I would magically give meteorologists a method to instantly simulate every atom of the atmosphere, they still couldn’t exactly tell me what the weather would be like in a week. It’s too dependent on small input changes for that.
> A rocketry analogy would be Archimedes dreaming about people traveling to the stars.
What’s wrong with that? The claim was „some day“, not in 5 years. It could take a few centuries or longer.
It could also be impossible, which for some reason people hardly consider.
It might be. But is there any hint that there are things we can absolutely not simulate?
I think there’s a good chance we won’t be able to simulate a exact propagation of a brain into the future due to quantum effects and unknowable starting states, but I don’t see how simulating one possible future could be impossible.
We have some hints to be skeptical on the prospect, I would say.
Like the fact we can't even explain an everyday phenomena all humans experience like consciousness. If you don't understand a common phenomena that (supposedly) occurs in the brain, you have a fundamental lack of knowledge on how the brain itself works, and thus simulating it would not be a thing possible to do.
Another hint is the fact that we can't currently even simulate a single cell because of how much complex they are, and the most advanced neuron models (like hodgkin-huxley) are still gross simplifications of how we think a real neuron works. We don't have any proof that this is a possible thing, it is something people believe it is the case, but pretty much could be a dead-end.
Other is that we don't have a way to reliably know the state of the alive brains without modifying them, so reproducing one with fidelity appears to be very hard or maybe even impossible.
And so on.
There's a genuine question of whether fully simulating a brain will be enough. We have several hundred million neurons in our digestive system. What we eat, and the kind of bacteria that lives there influences our mood. Same with the rest of our body. Brains are part of a larger organism. What would it mean to just simulate the brain independent of a body? Our sensory organs play a role in processing the incoming sensory stimulus and then send that off to the brain.
There's a good deal of sci-fi on the topic. My favorite of them is Neal Stephenson's Fall; or, Dodge in Hell.
For me it's Greg Egan's Permutation City.
That one's on my to-read list...
It appears inevitable that we can fully map a dead person's neurons and synapses. [1] is doing essentially that for a tiny sliver, with some amazing images to show. From there, it's "just" scaling up.
That alone wouldn't be enough to fully clone a person's consciousness. There is information stored in the actively firing synapses. For example short-term memory seems to be stored by sending signals in a loop, and there might be more such mechanisms. Those signals are obviously lost once the brain is dead. Another issue are hormones. The same brain regulated by a different (simulated) body might behave completely different. And then there are probably a lot of unknown unknowns. Despite decades of research there are still a lot of open questions, and more questions will become apparent once we actually start simulating complex brains.
But that doesn't mean that those early methods wouldn't be useful, both for science and for more questionable efforts. For example accessing the long-term memory of a recently deceased might be comparatively viable if given enough funding
1: https://edition.cnn.com/2024/05/15/world/human-brain-map-har...
> From there, it's "just" scaling up.
I suspect this wasn't your intention, but I feel this heavily undersells how much work is involved in "scaling up" to simulating a human brain. I wouldn't even say that it is inevitable, because there are so many unsolved questions and unknown-unknowns.
There are decades of research and we are still an unknown and large number of years away from doing this. Fusion power is more tractable that this.
It's not even clear whether our current approach to computation will ever be able to do this. We might need completely novel types of computers, maybe organic-machine hybrids.
I'm not even touching on the very real and serious ethical questions of simulating human level consciousnesses.
Hence the scare quotes around just.
Also note how my "just" only applies to scaling from mapping a grain-of-rice-sized piece of human brain to mapping a full human brain. Going from there to simulating it would be another big leap, never mind the challenges to simulate it in a way that actually produces results comparable to the actual brain of that person.
A human mind simulating another human mind is a computational system which is powerful enough to do arithmetic acting on itself, so Gödel's incompleteness theorems apply.
I’m not sure where you’re going with this.
To wit, no one expects human brains to be capable of arbitrarily complex computation.
What I'm getting at is that you won't achieve anything by simulating a human brain. Which doesn't mean nobody will try, the World's Richest Man seems determined to send humans to Mars, which is equally futile, but just that it is definitely futile and here's why we know that.
> you won't achieve anything by simulating a human brain
This does not follow from any theorem of Gödel.